IN VITRO vs. IN VIVO FUNCTION The most reliable approach for establishing the in vivo function of an enzyme is by genetics coupled with biochemical analyses. Numerous examples demonstrate that the physiological roles assigned to enzymes based on in vitro reactions they catalyze are incorrect. In perhaps tne most famous example, Arthur Kornberg discovered DNA polymerase I which was thought to be the sole E. coli DNA polymerase and, therefore, must be responsible for DNA replication [2]. However, mutants lacking DNA polymerase I grew and made DNA normally; hence, DNA polymerase I could not be the replicatlve polymerase [3, 4]. The replicatlve polymerase later was shown to be DNA polymerase 111, a much more complicated enzyme that, given the reaction conditions and template used for DNA polymerase I, is virtually inactive. However, mutants of polymerase I are super-sensitive to both UV and mutagens, showing that the in vivo role of polymerase I is DNA repair [5]. Kornberg desen/ed the Nobel Prize, but the fact remains that the physiological role of DNA polymerase I was in error due to the lack of genetics. A less famous but immediately relevant example of the in vivo ambiguity of an In vitro assigned function is provided by the computationally predicted and experimentally verified assignments of the N-succinyl Arg/Lys racemase [6] and L-Ala-D/L-Phe dipeptide epimerase [7] functions to members the MLE subgroup of the enolase superfamily. In both examples, Jacobson (Computation Core) predicted substrate promiscuity that was confirmed by enzymatic assays (EN Bridging Project). However, for both enzymes, the identity of the in vivo substrate as well as the physiological importance of the reaction is unknown. The Microbiology Core will apply genetic analyses and metabolomics to determine the in vivo roles of enzymes for which the In vitro functions are predicted by the Computation Core and verified by the Bridging Projects. Enzymes that emerge from the bottom of the funnel of Figure 1 in the Program Summary may be active with one, a few, or many related substrates with varying catalytic efficiencies. Which substrate(s) do they use in vivo? A combination of bacterial genetics, phenotypic characterization, and metabolite analysis will enable the physiological roles of the EFI targets to be evaluated and assigned.